https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113063
--- Comment #10 from GCC Commits <cvs-commit at gcc dot gnu.org> --- The master branch has been updated by Patrick Palka <ppa...@gcc.gnu.org>: https://gcc.gnu.org/g:37977343ff4f9dcb047d966d8cbaa222964763f9 commit r15-3491-g37977343ff4f9dcb047d966d8cbaa222964763f9 Author: Patrick Palka <ppa...@redhat.com> Date: Thu Sep 5 14:31:00 2024 -0400 c++: local class memfn synth from noexcept context [PR113063] Extending the PR113063 testcase to additionally constant evaluate the <=> expression causes us to trip over the assert in cxx_eval_call_expression /* We used to shortcut trivial constructor/op= here, but nowadays we can only get a trivial function here with -fno-elide-constructors. */ gcc_checking_assert (!trivial_fn_p (fun) || !flag_elide_constructors /* We don't elide constructors when processing a noexcept-expression. */ || cp_noexcept_operand); since the local class's <=> was first used and therefore synthesized in a noexcept context and so its definition contains unelided trivial constructors. This patch fixes this by clearing cp_noexcept_operand alongside cp_unevaluated_context in the function-local case of maybe_push_to_top_level. PR c++/113063 gcc/cp/ChangeLog: * name-lookup.cc (local_state_t): Clear and restore cp_noexcept_operand as well. gcc/testsuite/ChangeLog: * g++.dg/cpp2a/spaceship-synth16.C: Also constant evaluate the <=> expression. * g++.dg/cpp2a/spaceship-synth16a.C: Likewise. Reviewed-by: Jason Merrill <ja...@redhat.com>
